ABSTRACT

In this paper, we propose a side matching vector quantisation (SMVQ) data hiding scheme for image using multiple linear regression prediction. For each pixel block, the proposed scheme combines the multiple linear regression algorithm and the SMVQ algorithm, so that it can more accurately match the codeword or directly obtain the predicted value closer to the real pixel. Our experimental results show that when the VQ codebook sizes are 128, 256, 512, and 1024, and the SCB size is 16, this scheme obtains a better compression rate and information embedding ability. It can be concluded from the experimental results that this scheme is superior to existing algorithms in terms of compression and embedding capacity.

摘要

在本文中，我们提出一种用于图像的侧匹配矢量量化（SMVQ）数据隐藏方案使用多元线性回归预测。对于每个像素块，所提出的方案结合了多元线性回归算法和算法SMVQ，使得其可以更准确地匹配的码字，或直接获得的预测值接近真实像素。我们的实验结果表明，当VQ码大小为128，256，512，和1024，以及SCB大小为16，该方案获得更好的压缩率和信息嵌入能力。它可以从实验结果可以得出结论，该方案是优于在压缩方面现有的算法和嵌入容量。